Repertoires

Aug 25, 2026

CDRH3 is necessary but insufficient to guarantee binding to a particular epitope

Aug 19, 2026

Clonal expansion correlates with antigen binding

Jul 20, 2026

Different immune repertoires converge on the same V genes for specific antigens

antibodies/repertoires

Jul 20, 2026

Explicit germline encoding improves prediction of non-germline CDRH3 residues

Jul 20, 2026

Explicit germline encoding improves zero-shot antibody-affinity prediction

Apr 21, 2026

Antibodies from different repertoires within a given species targeting the same epitope often have the same light chain isotype and share critical residues

Apr 21, 2026

Allelic exclusion and inclusion

Apr 21, 2026

Antibody language models are able to distinguish correctly and incorrectly paired sequences, but are sensitive to scale and training dataset size

Apr 21, 2026

CDRH3 is shorter in mice and therapeutic Abs

Apr 21, 2026

Different antibody lineages can undergo convergent evolution during affinity maturation

Apr 21, 2026

GRAB motifs

Apr 21, 2026

Germline

antibodies/repertoires

Apr 21, 2026

Individual D genes are evolutionarily selected for well-defined antigen-binding capabilities

Apr 21, 2026

Light chain coherence

Apr 21, 2026

Next-generation sequencing

Apr 21, 2026

Observable Antibody Space

Apr 21, 2026

PLMs are biased toward germline antibodies

Apr 21, 2026

PLMs cluster antibodies from the same repertoire by V gene and maturation status

Apr 21, 2026

Somatic hypermutation

Apr 21, 2026

Some light chain V-genes are more likely to be found in dysfunctional antibodies

Apr 21, 2026

Sources of sequence diversity in antibodies

antibodies/repertoires

Apr 21, 2026

There are 4.6 million possible combinations of human antibodies given VDJ and VJ scrambling alone

antibodies/repertoires

Apr 21, 2026

Transgenic mice with only a single VH gene are nonetheless able to make effective antibodies against many targets

Apr 21, 2026

Germline usage determines whether nanobody CDR3 is kinked or extended

Apr 21, 2026

Individual V, D, and J gene usage is highly uneven

antibodies/repertoires

Apr 20, 2026

Immune repertoires

antibodies/repertoires

Quartz 4

Explorer

Repertoires

CDRH3 is necessary but insufficient to guarantee binding to a particular epitope

Clonal expansion correlates with antigen binding

Different immune repertoires converge on the same V genes for specific antigens

Explicit germline encoding improves prediction of non-germline CDRH3 residues

Explicit germline encoding improves zero-shot antibody-affinity prediction

Antibodies from different repertoires within a given species targeting the same epitope often have the same light chain isotype and share critical residues

Allelic exclusion and inclusion

Antibody language models are able to distinguish correctly and incorrectly paired sequences, but are sensitive to scale and training dataset size

CDRH3 is shorter in mice and therapeutic Abs

Different antibody lineages can undergo convergent evolution during affinity maturation

GRAB motifs

Germline

Individual D genes are evolutionarily selected for well-defined antigen-binding capabilities

Light chain coherence

Next-generation sequencing

Observable Antibody Space

PLMs are biased toward germline antibodies

PLMs cluster antibodies from the same repertoire by V gene and maturation status

Somatic hypermutation

Some light chain V-genes are more likely to be found in dysfunctional antibodies

Sources of sequence diversity in antibodies

There are 4.6 million possible combinations of human antibodies given VDJ and VJ scrambling alone

Transgenic mice with only a single VH gene are nonetheless able to make effective antibodies against many targets

Germline usage determines whether nanobody CDR3 is kinked or extended

Individual V, D, and J gene usage is highly uneven

Immune repertoires